Explore the vast range of titles available.

Australia has historically used structural descriptors of height and cover to characterize, differentiate, and map the distribution of woody vegetation across the continent but no national satellite-based structural classification has been available. In this study, we present a new 30-m spatial resolution reference map of Australian forest and woodland structure (height and cover), with this generated by integrating Landsat Thematic Mapper (TM) and Enhanced TM, Advanced Land Observing Satellite (ALOS) Phased Arrayed L-band Synthetic Aperture Radar (PALSAR) and Ice, Cloud, and land Elevation (ICESat),and Geoscience Laser Altimeter System (GLAS) data. ALOS PALSAR and Landsat-derived Foliage Projective Cover (FPC) were used to segment and classify the Australian landscape. Then, from intersecting ICESat waveform data, vertical foliage profiles and height metrics (e.g., 95% percentile height, mean height and the height to maximum vegetation density) were extracted for each of the classes generated. Within each class, and for selected areas, the variability in ICESat profiles was found to be similar with differences between segments of the same class attributed largely to clearance or disturbance events. ICESat metrics and profiles were then assigned to all remaining segments across Australia with the same class allocation. Validation against airborne LiDAR for a range of forest structural types indicated a high degree of correspondence in estimated height measures. On this basis, a map of vegetation height was generated at a national level and was combined with estimates of cover to produce a revised structural classification based on the scheme of the Australian National Vegetation Information System (NVIS). The benefits of integrating the three datasets for segmenting and classifying the landscape and retrieving biophysical attributes was highlighted with this leading the way for future mapping using ALOS-2 PALSAR-2, Landsat/Sentinel-2, Global Ecosystem Dynamics Investigation (GEDI), and ICESat-2 LiDAR data. The ability to map across large areas provides considerable benefits for quantifying carbon dynamics and informing on biodiversity metrics.

The aim of this research was to test recent developments in the use of Remotely Piloted Aircraft Systems or Unmanned Aerial Vehicles (UAV)/drones to map both pasture quantity as biomass yield and pasture quality as the proportions of key pasture nutrients, across a selected range of field sites throughout the rangelands of Queensland. Improved pasture management begins with an understanding of the state of the resource base, UAV based methods can potentially achieve this at improved spatial and temporal scales. This study developed machine learning based predictive models of both pasture measures. UAV-based structure from motion photogrammetry provided a measure of yield from overlapping high resolution visible colour imagery. Pasture nutrient composition was estimated from the spectral signatures of visible near infrared hyperspectral UAV sensing. An automated pasture height surface modelling technique was developed, tested and used along with field site measurements to predict further estimates across each field site. Both prior knowledge and automated predictive modelling techniques were employed to predict yield and nutrition. Pasture height surface modelling was assessed against field measurements using a rising plate meter, results reported correlation coefficients (R2) ranging from 0.2 to 0.4 for both woodland and grassland field sites. Accuracy of the predictive modelling was determined from further field measurements of yield and on average indicated an error of 0.8 t ha−1 in grasslands and 1.3 t ha−1 in mixed woodlands across both modelling approaches. Correlation analyses between measures of pasture quality, acid detergent fibre and crude protein (ADF, CP), and spectral reflectance data indicated the visible red (651 nm) and red-edge (759 nm) regions were highly correlated (ADF R2 = 0.9 and CP R2 = 0.5 mean values). These findings agreed with previous studies linking specific absorption features with grass chemical composition. These results conclude that the practical application of such techniques, to efficiently and accurately map pasture yield and quality, is possible at the field site scale; however, further research is needed, in particular further field sampling of both yield and nutrient elements across such a diverse landscape, with the potential to scale up to a satellite platform for broader scale monitoring.

Landsat 8 images have been widely used for many applications, but cloud and cloud-shadow cover issues remain. In this study, multitemporal cloud masking (MCM), designed to detect cloud and cloud-shadow for Landsat 8 in tropical environments, was improved for application in sub-tropical environments, with the greatest improvement in cloud masking. We added a haze optimized transformation (HOT) test and thermal band in the previous MCM algorithm to improve the algorithm in the detection of haze, thin-cirrus cloud, and thick cloud. We also improved the previous MCM in the detection of cloud-shadow by adding a blue band. In the visual assessment, the algorithm can detect a thick cloud, haze, thin-cirrus cloud, and cloud-shadow accurately. In the statistical assessment, the average user’s accuracy and producer’s accuracy of cloud masking results across the different land cover in the selected area was 98.03% and 98.98%, respectively. On the other hand, the average user’s accuracy and producer’s accuracy of cloud-shadow masking results was 97.97% and 96.66%, respectively. Compared to the Landsat 8 cloud cover assessment (L8 CCA) algorithm, MCM has better accuracies, especially in cloud-shadow masking. Our preliminary tests showed that the new MCM algorithm can detect cloud and cloud-shadow for Landsat 8 in a variety of environments.

When characterizing the processes that shape ecosystems, ecologists increasingly use the unique perspective offered by repeat observations of remotely sensed imagery. However, the concept of change embodied in much of the traditional remote-sensing literature was primarily limited to capturing large or extreme changes occurring in natural systems, omitting many more subtle processes of interest to ecologists. Recent technical advances have led to a fundamental shift toward an ecological view of change. Although this conceptual shift began with coarser-scale global imagery, it has now reached users of Landsat imagery, since these datasets have temporal and spatial characteristics appropriate to many ecological questions. We argue that this ecologically relevant perspective of change allows the novel characterization of important dynamic processes, including disturbances, long-term trends, cyclical functions, and feedbacks, and that these improvements are already facilitating our understanding of critical driving forces, such as climate change, ecological interactions, and economic pressures.

Data pre-processing for developing a generalised land use and land cover (LULC) deep learning model using earth observation data is important for the classification of a different date and/or sensor. However, it is unclear how to approach deep learning segmentation problems in earth observation data. In this paper, we trialled different methods of data preparation for Convolutional Neural Network (CNN) training and semantic segmentation of LULC features within aerial photography over the Wet Tropics and Atherton Tablelands, Queensland, Australia. This was conducted by trialling and ranking various training patch selection sampling strategies, patch and batch sizes, data augmentations and scaling and inference strategies. Our results showed: a stratified random sampling approach for producing training patches counteracted class imbalances; a smaller number of larger patches (small batch size) improves model accuracy; data augmentations and scaling are imperative in creating a generalised model able to accurately classify LULC features in imagery from a different date and sensor; and producing the output classification by averaging multiple grids of patches and three rotated versions of each patch produced a more accurate and aesthetic result. Combining the findings from the trials, we fully trained five models on the 2018 training image and applied the model to the 2015 test image. The output LULC classifications achieved an average kappa of 0.84, user accuracy of 0.81, and producer accuracy of 0.87. Future research using CNNs and earth observation data should implement the findings of this project to increase LULC model accuracy and transferability.

Forest structure attributes produced from terrestrial laser scanning (TLS) rely on normalisation of the point cloud values from sensor coordinates to height above ground. One method to do this is through the derivation of an accurate and repeatable digital elevation model (DEM) from the TLS point cloud that is used to adjust the height. The primary aim of this paper was to test a number of TLS scan configurations, filtering options and output DEM grid resolutions (from 0.02 m to 1.0 m) to define a best practice method for DEM generation in sub-tropical forest environments. The generated DEMs were compared to both total station (TS) spot heights and a 1-m DEM generated from airborne laser scanning (ALS) to assess accuracy. The comparison to TS spot heights found that a DEM produced using the minimum elevation (minimum Z value) from a point cloud derived from a single scan had mean errors >1 m for DEM grid resolutions <0.2 m at a 25-m plot radius. At a 1-m grid resolution, the mean error was 0.19 m. The addition of a filtering approach that combined a median filter with a progressive morphological filter and a global percentile filter was able to reduce mean error of the 0.02-m grid resolution DEM to 0.31 m at a 25-m plot radius using all returns. Using multiple scan positions to derive the DEM reduced the mean error for all DEM methods. Our results suggest that a simple minimum Z filtering DEM method using a single scan at the grid resolution of 1 m can produce mean errors <0.2 m, but for a small grid resolution, such as 0.02 m, a more complex filtering approach and multiple scan positions are required to reduced mean errors. The additional validation data provided by the 1-m ALS DEM showed that when using the combined filtering method on a point cloud derived from a single scan at the plot centre, errors between 0.1 and 0.5 m occurred in the TLS DEM for all tested grid resolutions at a plot radius of 25 m. These findings present a protocol for DEM production from TLS data at a range of grid resolutions and provide an overview of factors affecting DEMs produced from single and multiple TLS scan positions.

Fallows are widespread in dryland cropping systems. However, timely information about their spatial extent and location remains scarce. To overcome this lack of information, we propose to classify fractional cover data from Sentinel-2 with biased support vector machines. Fractional cover images describe the land surface in intuitive, biophysical terms, which reduces the spectral variability within the fallow class. Biased support vector machines are a type of one-class classifiers that require labelled data for the class of interest and unlabelled data for the other classes. They allow us to extrapolate in-situ observations collected during flowering to the rest of the growing season to generate large training data sets, thereby reducing the data collection requirements. We tested this approach to monitor fallows in the northern grains region of Australia and showed that the seasonal fallow extent can be mapped with >92% accuracy both during the summer and winter seasons. The summer fallow extent can be accurately mapped as early as mid-December (1–4 months before harvest). The winter fallow extent can be accurately mapped from mid-August (2–4 months before harvest). Our method also detected emergence dates successfully, indicating the near real-time accuracy of our method. We estimated that the extent of fallow fields across the northern grains region of Australia ranged between 50% in winter 2017 and 85% in winter 2019. Our method is scalable, sensor independent and economical to run. As such, it lays the foundations for reconstructing and monitoring the cropping dynamics in Australia.

The aim of this research is to expand recent developments in the mapping of pasture yield with remotely piloted aircraft systems to that of satellite-borne imagery. To date, spatially explicit and accurate information of the pasture resource base is needed for improved climate-adapted livestock rangeland grazing. This study developed deep learning predictive models of pasture yield, as total standing dry matter in tonnes per hectare (TSDM (tha−1)), from field measurements and both remotely piloted aircraft systems and satellite imagery. Repeated remotely piloted aircraft system structure measurements derived from structure from motion photogrammetry provided measures of pasture biomass from many overlapping high-resolution images. These measurements were taken throughout a growing season and were modelled with persistent photosynthetic pasture responses from various Planet Dove high spatial resolution satellite image-derived vegetation indices. Pasture height modelling as an input to the modelling of yield was assessed against terrestrial laser scanning and reported correlation coefficients (R2) from 0.3 to 0.8 for both a coastal grassland and inland woodland pasture. Accuracy of the predictive modelling from both the remotely piloted aircraft system and the Planet Dove satellite image estimates of pasture yield ranged from 0.8 to 1.8 TSDM (tha−1). These results indicated that the practical application of repeated remotely piloted aircraft system derived measures of pasture yield can, with some limitations, be scaled-up to satellite-borne imagery to provide more temporally and spatially explicit measures of the pasture resource base.

The emergence of cloud computing, big data analytics, and machine learning has catalysed the use of remote sensing technologies to enable more timely management of sustainability indicators, given the uncertainty of future climate conditions. Here, we examine the potential of “regenerative agriculture”, as an adaptive grazing management strategy to minimise bare ground exposure while improving pasture productivity. High-intensity sheep grazing treatments were conducted in small fields (less than 1 ha) for short durations (typically less than 1 day). Paddocks were subsequently spelled to allow pasture biomass recovery (treatments comprising 3, 6, 9, 12, and 15 months), with each compared with controls characterised by lighter stocking rates for longer periods (2000 DSE/ha). Pastures were composed of wallaby grass (Austrodanthonia species), kangaroo grass (Themeda triandra), Phalaris (Phalaris aquatica), and cocksfoot (Dactylis glomerata), and were destructively sampled to estimate total standing dry matter (TSDM), standing green biomass, standing dry biomass and trampled biomass. We invoked a machine learning model forced with Sentinel-2 imagery to quantify TSDM, standing green and dry biomass. Faced with La Nina conditions, regenerative grazing did not significantly impact pasture productivity, with all treatments showing similar TSDM, green biomass and recovery. However, regenerative treatments significantly impacted litterfall and trampled material, with high-intensity grazing treatments trampling more biomass, increasing litter, enhancing surface organic matter and decomposition rates thereof. Pasture digestibility and sward uniformity were greatest for treatments with minimal spelling (3 months), whereas both standing senescent and trampled material were greater for the 15-month spelling treatment. TSDM prognostics from machine learning were lower than measured TSDM, although predictions from the machine learning approach closely matched observed spatiotemporal variability within and across treatments. The root mean square error between the measured and modelled TSDM was 903 kg DM/ha, which was less than the variability measured in the field. We conclude that regenerative grazing with short recovery periods (3–6 months) was more conducive to increasing pasture production under high rainfall conditions, and we speculate that – in this environment - high-intensity grazing with 3-month spelling is likely to improve soil organic carbon through increased litterfall and trampling. Our study paves the way for using machine learning with satellite imagery to quantify pasture biomass at small scales, enabling the management of pastures within small fields from afar.

Ecological phenomena operate at different spatial scales and are not uniform across landscapes or through time. One ecological theory that attempts to account for scaling and spatiotemporal variances is hierarchical patch dynamics. It introduces a hierarchical patch network with smaller spatiotemporal scales being nested within larger scales. However, few studies have modeled its presence within animal population dynamics. Locusts are an excellent model for investigating the spatiotemporal hierarchy of animal population dynamics, due to their high migratory capacity, large geographic ranges that extend across widely differing environments, and available long‐term data on distributions. Here, we investigated the influence of preceding vegetation growth on desert locust Schistocerca gregaria and Australian plague locust Chortoicetes terminifera outbreaks on three spatial levels (species range > geographic region > land unit) and between seasons. Both species are dryland herbivores with population dynamics linked to habitat productivity pulses after rain. We used NDVI data (MODIS imagery) as a measure of vegetation growth in hierarchical generalized additive models at different scales. Locust outbreaks were either preceded by vegetation growth between 78 and 32 days (Australian plague locusts) or 32 and 20 days before (desert locust) the observation. Although prior vegetation growth characterized outbreaks of both species, the temporal pattern of NDVI differed between spatiotemporal levels. All model selection criteria selected for a similar spatial hierarchy for both species: geographic region > land unit which supports the hierarchical patch dynamics paradigm. Further, it illuminates important timing differences between geographic regions and land units for preceding vegetation growth and locust outbreaks which can help locust managers identify when and where outbreaks occur. By acknowledging the spatiotemporal patterning of locust abundance, we account for heterogeneity of population dynamics throughout species ranges. Our findings demonstrate the importance of incorporating spatiotemporal variation in population models of insects and other animals.